Loss of function of Ribonuclease T2, an ancient and phylogenetically conserved RNase, plays a crucial role in ovarian tumorigenesis

Francesco Acquatia, Marta Lualdia, Sabrina Bertilacciob, Laura Montia, Giovanna Turconia, Marco Fabbric, Annalisa Grimaldid, Achille Anselmoe, Antonio Inforzatoe, Angelo Collottac, Laura Cimettif, Cristina Rivaf, Laura Gribaldoc, Paolo Ghiab, and Roberto Taramellia,1

Departments of aTheoretical and Applied Sciences and dBiotechnology and Life Sciences, Università degli Studi dell’Insubria, 21100 Varese, Italy; bDepartment of Onco-Hematology, Clinical Unit of Lymphoid Malignancies, San Raffaele Scientific Institute, 20132 Milan, Italy; cMolecular Biology and Genomics Unit, Institute for Health and Consumer Protection, Joint Research Centre, 21027 Ispra, Italy; eClinical and Research Institute Humanitas, 20089 Rozzano, Italy; and fDepartment of Pathology, Ospedale di Circolo, 21100, Varese, Italy

Edited* by George Klein, Karolinska Institutet, Stockholm, Sweden, and approved April 3, 2013 (received for review December 18, 2012) In recent years, the role played by the stromal microenvironment The tumor microenvironment has recently become a crucial has been given growing attention in order to achieve a full under- player in the carcinogenetic processes (6–9). Indeed, stromal standing of cancer initiation and progression. Because cancer is a components such as fibroblasts, immune/vascular cells, and ECM tissue-based disease, the integrity of tissue architecture is a major are all known to be engaged in a complex and dynamic cross-talk constraint toward cancer growth. Indeed, a large contribution of the with the neighboring epithelium, which can crucially contribute natural resistance to cancer stems from stromal microenvironment to tumor suppression and resistance (6–9). Indeed, interaction of components, the dysregulation of which can facilitate cancer occur- parenchymal cells with stromal components is known to be es- rence. For instance, recent experimental evidence has highlighted sential for the maintenance of tissue architecture and function the involvement of stromal cells in ovarian carcinogenesis, as epit- (6–9), and several studies have documented the role played by omized by ovarian xenografts obtained by a double KO of the mu- the stroma as a key determinant of epithelial proliferation, cell GENETICS Dicer Pten rine and . Likewise, we reported the role of an death, motility, polarity, and differentiation (6, 9). Moreover, ancient extracellular RNase, called Ribonuclease T2 (RNASET2), stromal abnormalities have long been known to exert profound within the ovarian stromal microenvironment. Indeed, hyperexpres- influences on tumor induction and progression (6–9). However, RNASET2 sion of is able to control tumorigenesis by recruiting mac- although the notion of a dominant role of the microenvironment rophages (mostly of the anticancer M1 subtype) at the tumor sites. in providing signaling cues to suppress or revert a malignant We present biological data obtained by RNASET2 silencing in the – RNASET2 phenotype is consolidated (6 9), a detailed understanding of the poorly tumorigenetic and highly -expressing human types and specific roles of stromal and immune cells and of the OVCAR3 cell line. RNASET2 knockdown was shown to stimulate in ECM dynamics within it is still lacking. vivo tumor growth early after microinjection of OVCAR3 cells in Our research group has recently contributed to this issue by in- nude mice. Moreover, we have investigated by molecular profiling vestigating the role of the human RNASET2 (10–15). The the in vivo expression signature of human and mouse cell xeno- latter codes for an extracellular RNase expressed in several tissues, grafts and disclosed the activation of pathways related to activation of the innate immune response and modulation of ECM compo- including the ovary and tubal structures, and is highly conserved nents. Finally, we provide evidence for a role of RNASET2 in trigger- among phyla from viruses to humans, implying a crucial, evolu- ing an in vitro chemotactic response in macrophages. These results tionary conserved function (16). In an ovarian cancer xenograft further highlight the critical role played by the microenvironment in model, we previously found RNASET2 to carry out a strong onco- RNASET2-mediated ovarian tumor suppression, which could even- suppressive activity associated with an in vivo recruitment of mac- tually contribute to better clarify the pathogenesis of this disease. rophages into the tumor mass (15). Here, we report RNASET2 gene silencing in an independent ovarian cancer cell model to provide further evidence for its role in microenvironmental-mediated tumor espite its relatively low incidence rate, ovarian cancer is an suppression in vivo. Dextremely lethal disease, which globally claims 125,000 lives per year, making it the seventh leading cause of cancer-related Results deaths among women (1). Although the 5-y survival rate for stage I RNASET2 Silencing in OVCAR3 Ovarian Cancer Cells Is Associated with ovarian cancer is >90%, stage I diagnoses are more often the Increased in Vivo Tumorigenesis. Ovarian cancer-derived OVCAR3 exception than the rule. Indeed, most patients present with cells were used to investigate the effects of RNASET2 knock- advanced-stage tumors, for which the 5-y survival rate is a dismal down. This cell line was chosen on two main basic premises. 30% (1). Contributing to such a dramatic outlook is the rather poor First, RNASET2 transcript and protein levels are significantly understanding of the biological basis of the disease, which appears higher compared with other ovarian cancer cell lines (10). as a highly heterogeneous group of pathological entities (2, 3). Traditionally, the anatomical site from which most cases of ovarian cancer were thought to arise has been the ovarian surface epithe- Author contributions: F.A. and R.T. designed research; M.L., S.B., L.M., G.T., M.F., A.G., A.A., lium (2, 3). This long-held belief has recently been challenged by A.I., A.C., L.C., C.R., and L.G. performed research; F.A., C.R., L.G., P.G., and R.T. analyzed several observations pointing to fallopian tube structures as the data; and F.A. and R.T. wrote the paper. prevailing sites where insurgence of ovarian neoplasias occurs (4). The authors declare no conflict of interest. According to this new paradigm, the ovary would rather represent *This Direct Submission article had a prearranged editor. “ ” a sort of metastatic secondary site, the primary site being local- Data deposition: The data reported in this paper have been deposited in the Gene Ex- ized in the fallopian tube parenchyma (4). Supporting this para- pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE43029). digm is a recent finding from Dicer/Pten double KO mice, in which 1To whom correspondence should be addressed. E-mail: [email protected]. ovarian carcinogenesis is initiated from abnormal proliferation This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. within the stromal compartment of the fallopian tube (5). 1073/pnas.1222079110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1222079110 PNAS Early Edition | 1of6 Downloaded by guest on September 24, 2021 Second, OVCAR3 cells are reported to be poorly tumorigenic in in control tumors, that was already observed in histological sec- vivo (17). RNASET2 silencing was carried out in OVCAR3 cells tions, was also confirmed by IHC analysis (Fig. S1). by stably expressing an siRNA targeting the RNASET2 mRNA, A relevant histological finding concerned the presence of in- resulting in complete abrogation of its expression (Fig. 1A). flammatory infiltrates in xenografts from control OVCAR3 cells. Three independent RNASET2 knockdown clones and three To characterize the nature of such infiltrating cells, we carried out control clones (expressing a scrambled shRNA) were then cho- IHC assays for several inflammatory markers (data not shown). sen for further analyses. The clones were injected s.c. into nude Among the chosen markers the most significant was CD68, whose mice, and tumor growth was followed for approximately 40 d. As pattern pointed to the presence of murine macrophages as the most depicted in Fig. 1B, after a few days of in vivo growth, the three representative cell population of the infiltrate, as shown in Fig. 2 G RNASET2-silenced clones showed an increased growth kinetics and H. To discriminate between macrophage subpopulations, we compared with control clones. By the end of the in vivo assay, the carried out an IHC assay for markers able to differentiate the M1 average tumor weight from RNASET2-silenced xenografts was and M2 macrophage cell subpopulations. As shown in Fig. 2 I and also significantly increased compared with control xenografts J, a consistent depletion of inducible nitric oxide synthase-positive (Fig. 1C). We thus concluded that RNASET2 silencing conferred M1 macrophages was observed in RNASET2-silenced tumors com- to the poorly tumorigenic OVCAR3 cell line an increased in vivo pared with control xenografts. By contrast, infiltration by dectin-1– growth rate. Such behavior is substantially congruent with what positive M2 macrophages, which was less pronounced in control we have previously reported following overexpression of RNASET2 xenografts compared with M1 macrophages, was only slightly de- into the highly tumorigenic, low RNASET2-expressing Hey3Met2 creased in RNASET2-silenced tumors (Fig. 2 K and L). In keeping ovarian cancer cell line, as, in that case, tumorigenesis was strongly with these results, a subsequent analysis of OVCAR3-derived xe- inhibited by overexpression of this gene (15). nograft RNA by real-time quantitative PCR (qPCR) technique using primer pairs specific for murine markers of M1/M2 polari- Morphological Characterization of RNASET2-Silenced OVCAR3 Cell zation showed a decreased expression for two of three M1-specific Tumors Grown in Vivo. To compare the morphological features of markers in RNASET2-silenced xenografts. By contrast, no effect RNASET2-silenced and control xenografts, tumor samples were on gene expression was observed for two M2-specific markers, excised and processed for histological and immunohistochemical whereas expression of the third marked tested was up-regulated in (IHC) analyses (Table S1). As depicted in Fig. 2, all tumors from RNASET2-silenced xenografts (Fig. S2). RNASET2-silenced OVCAR3 cells were characterized by a diffuse Together, these results are totally in keeping with our previous pattern (Fig. 2 A and C). Tumor cells were large and polygonal, data from the Hey3Met2 ovarian model (15), in which a strong with prominent nuclei and remarkable nucleoli. Significantly, infiltration of inflammatory cells (mostly M1 macrophages) was atypical mitotic figures were frequent, and the mitotic and pro- observed in RNASET2-overexpressing tumor xenografts (15). liferative indexes were high (Fig. 2E and Table S1). By contrast, The results presented here thus provide further support to the tumors from control OVCAR3 cells showed a widespread nests/ notion that RNASET2-mediated tumor suppression is associated micronests pattern, with focal fibrous septa (Fig. 2 B and D). with in vivo recruitment of the M1 subclass of macrophages, Coagulative necrosis and apoptosis in central areas of the tumors which is known to have antitumorigenic properties (18). were abundantly present, whereas mitotic and proliferative indexes were decreased (Fig. 2F and Table S1). A more differentiated Gene Expression Profile of RNASET2-Silenced OVCAR3 Cell Tumor pattern, documented by increased cytokeratin 7 staining, was also Xenografts. To investigate the molecular pathways underlying observed in control samples compared with RNASET2-silenced RNASET2-mediated tumor suppression in vivo, we next investi- tumors (Fig. S1). The occurrence of areas of widespread apoptosis gated the gene expression profile of the tumor xenografts. To

Fig. 1. RNASET2 silencing enhances tumor growth in vivo. (A) Characterization of RNASET2-silenced OVCAR3 cells by Western blot analysis. (Upper) Intracellular lysates were probed with an anti-RNASET2 polyclonal antibody. (Lower) The same blot was probed with an anti–α-tubulin monoclonal antibody for normalization. (B) Three control and three RNASET2-silenced OVCAR3 clones were inoculated s.c. in nude mice as described in Materials and Methods. Tumor growth was monitored every 2 d until day 40. At least five mice were inoculated for each tested clone. Bars represent SD values (*P < 0.5 and **P < 0.01, significant difference in tumor size be- tween RNASET2-silenced and control tumors, Student t test). (C) The mean weights of the collected tumors are represented (P = 0.004, two-tailed Student t test).

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1222079110 Acquati et al. Downloaded by guest on September 24, 2021 RNASET2-knockdown Control Searches were then made for significantly enriched gene classes, xenogras xenogras as defined by (GO) annotation and Kyoto Ency- clopedia of Genes and Genomes (KEGG) analysis. As shown in AB Fig. S3A and Dataset S3, RNASET2-regulated human genes turned out to be significantly enriched in GO biological process H & E classes mainly related to leukocyte activation (12 genes) and development (53 genes). Significantly, most human genes related to leukocytes activation turned out to be down-regulated in RNASET2-silenced OVCAR3 cells, suggesting a positive regu- latory role for cells of the immune/inflammatory system by CD RNASET2. On the contrary, KEGG analysis displayed enrich- H & E ment in pathways mainly related to immunity, cell adhesion, and cancer. To our surprise, GO and KEGG analysis did not show a significant enrichment in terms related to apoptosis. This was an unexpected finding, considering the increased apoptotic rate that we observed by IHC analysis in control OVCAR3 cell- EF derived xenografts (Fig. S1). However, a closer inspection of the microarray data showed a few apoptosis-related genes whose Ki-67 expression was indeed increased in control xenografts compared with RNASET2-silenced tumors. We thus confirmed by the more sensitive real-time qPCR technique that expression of five proa- poptotic genes [Galanin (GAL), anterior pharynx defective 1 homolog B(APH1B), tumor necrosis factor receptor superfamily, member GH 25 (TNFRSF25), unc-13 homolog D (UNC13-D),andnuclear protein transcriptional regulator 1 (NUPR1)] and two antiapoptotic genes CD68 [adhesion molecule with Ig like domain 1 (AMIGO) and annexin

A4 (ANXA4)] was down-regulated and up-regulated, respectively, GENETICS in RNASET2-silenced xenografts. Thus, although not detected by microarray analysis, these genes might play a role in the decreased apoptotic rate observed in RNASET2-silenced xenografts. IJ When we turned to the host stromal transcriptional profile of the xenografts, a GO search returned gene classes related to cell ad- iNOS hesion and, most importantly, to immune response (Fig. S3B). The annotated murine genes in these two GO classes are listed in Dataset S4. Interestingly, similar results were obtained by KEGG analysis, with enrichment in pathways related to immunity and cell adhesion. As already reported for the analysis of the human cancer KL cell transcriptome, most murine genes related to immune system roles turned out to be down-regulated in RNASET2-silenced Decn-1 OVCAR3 cells, again suggesting a positive regulatory role for RNASET2 within the context of cells of the immune/inflammatory system. The significant enrichment of genes involved in immuno- logical functions is in keeping with recent data pointing to a role for Fig. 2. Histological and immunohistochemical analysis of tumor samples. members of the T2 RNase gene family in the immune response Parental and RNASET2-silenced OVCAR3-derived tumor sections were stained (Discussion). Being aware of the sensitivity limits of the microarray with H&E (A–D). Microphotographs show a prevailing diffuse pattern and hybridization technology, we aimed to further investigate whether some solid nests in the central areas of RNASET2-silenced tumors, filled by large RNASET2 might induce other immunologically competent genes. and polygonal cells with prominent nuclei and nucleoli (A and C). By contrast, We thus compared by qPCR control and RNASET2-silenced tumor control tumors show widespread nests/micronests pattern, with focal slender fi- xenografts for expression levels of 70 genes belonging to the mouse brous septa (B and D). Tumor sections were also analyzed by IHC for Ki-67 as chemokine gene family. Strikingly, 35 additional genes were shown a marker for cell proliferation (E and F)andforCD68(G and H), inducible Nitric to be significantly down-regulated in RNASET2-silenced tumors in Oxide Synthase (iNOS-1) (I and J), and dectin-1 (K and L)todetectmurinehost inflammatory cells, the whole macrophage cell population, and M1- and M2-po- addition to the 11 genes already detected by microarray analysis larized macrophages, respectively. Photomicrographs shown are representative (Dataset S5). Among the 35 gene entries, we found several genes images of multiple fields examined in at least four sections derived from tumor with an established role in monocyte/macrophage chemotaxis, such excised in two to three independent animals for each experimental group. as chemokine (C-C motif) ligand (Ccl)-2, Ccl-7, chemokine (C-C motif) receptor (Ccr)-2, Ccr-6, Ccr-7, Cx3cl-1,andCxcl-12,sug- gesting that RNASET2 might act in vivo to trigger a consistent evaluate the contribution of human cancer cells and murine chemoattractive response in cells of the innate immune system. host stroma, total RNA extracted from tumor xenografts was fluorescently labeled and hybridized to human and mouse whole- Analysis of RNASET2 as Candidate Chemotactic Protein. The results genome microarrays. By using a fold induction or repression greater reported here earlier strongly suggest that RNASET2 might carry than 1.5 and a P value lower than 0.01 as inclusion criteria, 104 out its tumor-suppressive role by a direct recruitment of host mac- human genes were found to be up-regulated and 189 were down- rophage cells into the tumor mass. We thus addressed whether a di- regulated in RNASET2-silenced xenografts (Dataset S1).When rect interaction of RNASET2 with cells of this lineage could the murine host stromal transcriptional profile was investigated, take place. 98 murine genes were shown to be up-regulated in RNASET2- To this end, we first investigated the occurrence of binding of silenced xenografts, compared with 272 down-regulated genes recombinant RNASET2 to the cell surface of human macrophages. (Dataset S2). For this purpose, human monocyte-derived macrophages upon in

Acquati et al. PNAS Early Edition | 3of6 Downloaded by guest on September 24, 2021 vitro differentiation (i.e., M0) and M1/M2 polarized ones were RNASET2 protein was largely impaired (Fig. 3B). These results used. After exposure to endotoxin-free recombinant RNASET2 are consistent with the hypothesis of a receptor-mediated action protein purified from Pichia pastoris,aflow cytometric analysis of extracellular RNASET2. showed significant and dose-dependent RNASET2 binding to the cell surface in all macrophage subpopulations tested (Fig. 3A). Discussion To investigate whether this binding could induce a cellular re- In the present study, we have extended some basic assumptions sponse, recombinant RNASET2 was used for in vitro chemotaxis concerning the role of RNASET2 with regard to some of its assays on a human promyelocytic subclone showing low endoge- pleiomorphic features, in particular those related to the control nous RNASET2 expression levels. Interestingly, we observed of cancer initiation. We have previously demonstrated that this a significant RNASET2-mediated increase in the migration rate of RNase has the capacity to control ovarian tumorigenesis in vivo these cells (Fig. 3 B and C). Importantly, such response was not (15). Indeed, induced overexpression of RNASET2 was coupled affected by the expression system chosen for recombinant protein with a strong reduction of ovarian tumorigenesis (15). The mech- expression, as RNASET2 produced with baculovirus/insect cells anism through which RNASET2 is acting was shown to take place and Pichia pastoris expression systems performed equally well in the in the microenvironment and consisted mainly of recruitment of chemotaxis assay. Moreover, the same chemotactic response was M1-polarized host macrophages toward the site of tumor growth. also observed when recombinant RNASET2 was applied to freshly Of note, this macrophage subpopulation is known to be associated prepared human primary monocytes isolated from buffy coats. with inhibition/suppression of cancer cells growth (19, 20). To gain Taken together, these data support the possibility that the more insights into the antineoplastic features of RNASET2, we RNASET2 protein might directly evoke a chemotactic response in reasoned that silencing its expression in ovarian cancer cell lines this cell lineage. This allowed us to hypothesize a receptor-medi- showing high endogenous levels of this protein would possibly be ated binding of the RNASET2 protein to the target cell surface. associated with the appearance of a cancer phenotype. Indeed, Intracellular signaling pathways in response to inflammatory che- following their inoculation into nude mice, RNASET2-silenced mokines are often triggered by the binding of signaling molecules OVCAR3 cells displayed completely different growth kinetics with to a cell-surface G protein-coupled receptor (19). Strikingly, when respect to control OVCAR3 cells, with the appearance of con- the in vitro migration assays on U937 cl.10 cells were carried out spicuous tumor masses just a few days after inoculation. following pretreatment with pertussis toxin (PTx), a known in- A closer examination of tumor sections revealed, among other hibitor of G protein-coupled receptor function, cell migration in re- distinctive features, a clear effect of RNASET2 silencing on the sponse to chemokine (C-C motif) ligand 2 (CCL-2) and recombinant occurrence of stromal infiltrates. Indeed, comparative IHC analysis

A

BC

Ctrl

RNASET2

PTx + RNASET2

Fig. 3. Binding of recombinant RNASET2 to the cell surface of macrophages and chemotactic activity on U937 cl.10 cell line. (A) Peripheral blood leucocyte (PBL)-derived human macrophages were detached from culture plates and starved in complete medium plus 1% FCS for 3 h at 37 °C. Binding assay was performed by adding recombinant RNASET2 for 3 h at 4 °C. Fc receptors were blocked by using 10% human serum for 15 min at 4 °C. Rabbit anti-human RNASET2 and normal rabbit IgG (as isotype control) were used to perform cell surface staining assay. Data are expressed as fold increase of mean fluorescence intensity (MFI) of anti- RNASET2 antibody-stained samples vs. isotype control-stained ones (error bars represent SEM). (B) U937 cl.10 cells were pretreated with 1 μg/mL PTx or left untreated. Human recombinant RNASET2 and CCL2 proteins were used as chemoattractants at the final concentrations of 8.172 μg/mL and 100 ng/mL, respectively. Statistical analysis was performed with the unilateral Student t test. (C) Fluorescence microscopy images of filters. Nuclei were stained with DAPI (magnification of 40×).

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1222079110 Acquati et al. Downloaded by guest on September 24, 2021 of the xenograft infiltrates revealed a substantial depletion of M1 that are down-regulated in RNASET2-silenced ovarian xenografts. macrophages in the RNASET2-silenced xenografts, suggesting that Several collagen genes and some matrix-remodeling enzymes were the microenvironment of control xenografts is consistently more included in this list as well. Clearly, the observed down-regulation tumor-inhibiting than the one in which RNASET2 is knocked down. of these genes in the RNASET2-null xenografts is expected to To provide molecular correlates of the RNASET2 induced bi- affect the stromal-ECM compartment architectural features, ological responses, we looked at the in vivo gene expression pro- which in turn would reorganize and impact the surrounding mi- files from xenograft samples. In this context, the results concerning croenvironment. Further insights into RNASET2 stromal func- the expression profile changes in the murine host stromal com- tions are also suggested by the observed down-regulation of partment are quite revealing. In general terms, our results sub- several cell-adhesion molecules. Involvement in and modulation stantiate the conclusions drawn from previous experimental data, of cell-to-cell contacts seems therefore to be a prerogative of which strongly suggest a role for RNASET2 in the innate immune RNASET2 function. Worth mentioning are some cytoskeletal and response (16, 21, 22). In fact, members of the T2 RNase protein cytoskeleton-associated genes that were also disclosed by GO family were recently shown to modulate the mammalian innate analysis under the term “development” and whose expression immune system, as epitomized by the omega-1 protein secreted turned out to be affected in RNASET2-silenced cells, suggesting from Schistosoma mansoni, which is able to prime human and that they could possibly impact the in vivo cell differentiation mouse dendritic cells for Th2 polarization of T cells (21, 22). pattern of the tumor cells, as disclosed in our xenograft model. Furthermore, a recent work from our group reported that the Thus, the gene expression profileswehavedocumentedinvivoin tumor suppressive role carried out by the RNASET2 gene in the our ovarian model system are basically in agreement with the Hey3Met2 ovarian cancer cell line model depends on the in vivo apparent functional behavior endorsed by RNASET2 that has recruitment of cells from the macrophage lineage (15). Finally, been proposed before now (16, 21–23). We emphasize its role in a potential link between T2 ribonuclease-mediated modulation of creating an environmental milieu that is basically tumor suppres- the immune system and tumor suppression was recently estab- sive. The latter point is not irrelevant considering that some lished by a report showing that the Human T-lymphotropic virus 1 RNases are now in phase III clinical trials (25). (HTLV-1) tax gene drives a strong down-regulation of the human As a last point, we would like to comment on a recent finding RNASET2 gene (23). These experimental evidences, which col- suggesting the fallopian tube as a potential primary site of or- lectively point to a role for RNASET2 in host immune response, igin of high-grade serous ovarian carcinomas (5). More specifi-

are quite well recapitulated in the experimental data presented cally, Kin et al. were recently able to demonstrate that the onset GENETICS here. Indeed, not only is RNASET2 knock-down in OVCAR3 cells of the carcinogenetic process takes place in the stromal com- associated with increased tumor growth in vivo, but we could also partment of the fallopian tube in conditional Dicer/Pten double- detect the occurrence of a significant decrease of M1-polarized knockdown mice (5). Intriguingly, these structures correspond to macrophage infiltration as a distinct feature of the RNASET2- sites of strong RNASET2 expression, suggesting that this extra- silenced tumors. Within this frame, the in vivo mouse stromal gene cellular RNase might be, among other players, a good candidate expression profile provided a strong support to this hypothesis, as for initiating and sustaining ovarian tumorigenesis. shown by the results of a GO search, which reported a significant In summary, we have investigated the putative mechanisms increase in categories such as “immune response,”“defense through which an extracellular RNase is involved in the control response,”“innate immune response,” and “acute inflammatory of ovarian tumorigenesis. The pleiomorphic functions of RNA- response.” Aconfirmation of the role played by RNASET2 in SET2 and the pathways that have been uncovered in the present modulating the innate immune response was also provided by study will likely contribute to shed some light on the dominant a KEGG search, which showed a significant enrichment for the antitumoral role exhibited by the stromal/microenvironmental terms “chemokine signaling” (13 genes), “cytokine-cytokine re- architectural determinants in ovarian cancer pathogenesis (6–9). ceptor interaction” (12 genes), and “leukocyte transendothelial mi- Furthermore, interindividual genetic variation within all the mo- gration” (10 genes). A subsequent analysis by real-time PCR assay lecular components comprising the networks where RNASET2 is further showed that an increased number of mouse chemokine genes likely to act might be exploited to better define the genetic basis of was actually affected in tumor xenografts following RNASET2 si- the cancer-resistance phenotype (7). lencing. Finally, analysis of the in vivo expression profile for human genes in RNASET2-silenced xenografts reported the involvement of Materials and Methods immune system-related pathways as well, such as “leukocyte acti- RNASET2 Knockdown in OVCAR3 Cells. RNASET2 silencing was carried out by vation” (12 genes) and “lymphocyte proliferation” (seven genes). means of a lentiviral infection vector expressing an shRNA targeting nucleotides Taken together, these data strongly support the occurrence of 814 to 832 within the RNASET2 mRNA (Genbank accession no. NM_003730.4). an RNASET2-mediated tumor suppressive mechanism based on Oligonucleotide sequences for RNAi were generated with the pSicoOligomaker the establishment of cross-talk between cancer cells and cellular/ 1.5 software and subsequently aligned to the sequence (UCSC fi structural components of the microenvironment, which results in Genome Browser) to test for speci city. Control scrambled sequences were an innate immune response against neoplastic cells. The basis of designed having the same length and GC content as the test sequence, but with no significant match to the database. The sequences of the RNASET2-targeting this presumed mechanism is being investigated by our group, and and scrambled sequences are available on request. Sense and antisense oligo- we started by demonstrating here that the RNASET2 protein nucleotides representing these sequences were annealed and cloned into HpaI/ actually displays chemotactic properties in vitro toward macro- XhoI-digested pSicoR vector (provided by F. Nicassio, Fondazione Istituto FIRC di phage cells. Viewed from this perspective, these data allow us to Oncologia Molecolare, Milan, Italy). The recombinant shRNA-expressing vectors hypothesize a role for RNASET2 similar to that of the alarmins were stably transfected into OVCAR3 cells with Lipofectamine 2000 (Invi- (24), although we have to provide further experimental evidence trogen). The efficiency of RNASET2 knockdown was evaluated by Western for that. In this context, worth mentioning are our preliminary blot analysis as described previously (14). data indicating that RNASET2 protein expression levels are rather high in ovarian cancer cells from low-grade human ovarian In Vivo Xenograft Studies. Male nude mice 6 wk of age were challenged s.c. in fl × 6 tumors but progressively decrease in higher-grade/stage tumor the left ank with 2 10 OVCAR3 cells in 0.15 mL saline solution. Animals were monitored twice per week for weight and tumor growth, and killed at samples. Concomitant with the immunological-based responses day 39. Separate experiments were performed with three different clones of mediated by RNASET2, in this work, we also detected other both control and RNASET2-silenced cells (n ≥ 5 animals per clone). Animals mouse stromal gene expression profile changes that are mainly were maintained in a pathogen-free facility and treated in accordance with related to tissue structural/architectural features. Indeed, GO and European Union guidelines under the approval of the ethical committee KEGG analyses have highlighted several ECM component genes of Ospedale San Raffaele (Institutional Animal Care and Use Committee

Acquati et al. PNAS Early Edition | 5of6 Downloaded by guest on September 24, 2021 protocol 418). Tumors were frozen for RNA extraction and microarray region of the mRNA used to design the probes for microarray experiments. studies. Statistical significance of data was calculated by Student t test. Real time RT-PCR reactions were performed on ABI PRISM 7000 system (Applied Biosystems) in triplicate in a 25 μL volume containing target cDNA, 40 nM Histological and IHC Assays. A total of nine formalin-fixed and paraffin- primers, 12.5 μL of Power SYBR Green Master Mix, and water. Samples were embedded tissue specimens from murine tumors were cut at 3 μm thickness denatured at 95 °C for 15 s and annealed/extended at 60 °C for 1 min, for 40 and stained with H&E for routine histopathological analysis. The following cycles. Fluorescent signals generated during PCR amplification were moni- parameters were evaluated by a senior pathologist (C.R.): type of growth, cell tored and analyzed with ABI PRISM 7000 SDS software (Applied Biosystems). morphology, stroma infiltration, necrosis, apoptosis, and mitotic index. For IHC Comparison of the amount of each gene transcript among different samples assays, formalin-fixed and paraffin-embedded additional sections were used. was made by using tubulin or GAPDH as reference. The amount of target Sections were mounted on slides, deparaffinized and immersed in 10 mM RNA, normalized to the endogenous reference gene, was calculated by means sodium citrate, pH 6, and microwaved for antigen retrieval. Antibodies raised of the difference in threshold cycle parameter (ΔΔCt). Statistical analysis was against the following markers were used: Ki-67 (Dako), CD68 (Dako), CD45RO done by Student t test. All primer sequences are available on request. (Dako), cytokeratin 7 (Ventana), anti-cleaved caspase-3 (Cell Signaling Tech- nology), inducible nitric oxide synthase (Transduction Lab), and dectin-1 (Bio- In Vitro Assays. Macrophages were obtained from buffy coat-derived human orbyt). Reactions were visualized by using an ABC immunoperoxidase system monocytes upon macrophage colony-stimulating factor (100 ng/mL) stimu- fi (Vector Labs) or UltraVision (Thermo Scienti c) with DAB (Sigma) as chromo- lation. LPS (100 ng/mL) plus IFN-γ (20 ng/mL) or IL-4 (20 ng/mL) was used genic substrate. The extent of immunoreactivity in the samples was assessed by for M1 and M2 differentiation, respectively. Macrophages were starved in a senior pathologist considering the proportion of positive cells. complete medium plus 1% FCS for 2 h at 37 °C before the binding assay, which was performed by adding recombinant RNASET2 for 3 h at 4 °C. Gene Expression Profiling. For microarray hybridization, RNA from tumor Fc receptors were blocked by using 10% (vol/vol) human serum. Rabbit samples or cultured cells was purified and quantified by UV spectrophotom- anti-human RNASET2 and normal rabbit IgG (as isotype control) were used etry. RNA quality was assessed with a model 2100 Bioanalyzer (Agilent). Cy3- to perform cell surface staining assay. Data were expressed as fold increase labeled cRNA was generated from 500 ng input total RNA. For every sample, of mean fluorescence intensity of RNASET2-stained samples vs. isotype 1.65 μg labeled cRNA was hybridized to the SurePrint G3 Human or Mouse control-stained ones. Gene Expression 8 × 60K Microarray (Agilent). The fluorescence intensities on The U937 cl.10 cell line (provided by R. Accolla) is an HIV-permissive sub- scanned images were extracted and preprocessed by Agilent Feature Ex- traction Software (version 10.5.1.1). Quality control and array normaliza- clone derived from the human histiocytic lymphoma-derived U937. Cells were tion was performed in the R statistical environment by using the Agi4 × grown in RPMI/10% (vol/vol) FBS with 1% L-glutamine. The in vitro migration 44PreProcess package downloaded from the Bioconductor Web site (www. rate of U937 cl.10 cells was assessed in blind-well chambers (NeuroProbe) with μ fi bioconductor.org). The quantile normalization and default filtering steps 3- m-pore polycarbonate lters separating the upper and lower chambers. were based on those described in the Agi4×44 PreProcess reference manual. Cells were resuspended in serum-free RPMI-1640 medium and incubated for 24 To detect expression differences among different cell populations hat37°C,5%CO2. PTx-pretreated and control U937 cl.10 cells were resus- 5 a moderated t test was applied. Moderated t statistics were generated by pended in serum-free medium at a final density of 6 × 10 cells/mL and placed Limma Bioconductor package, and modulated genes were chosen as those in the upper compartment. Chemoattractant agents (8,172 μg/mL RNASET2; with a fold change greater than 1.5 and a P value lower than 0.01. 100 ng/mL CCL2) were diluted in serum-free medium and added to the lower These genes were analyzed by using the WebGestalt Web site (http:// compartment. When reported, pretreatment with 1 μg/mL PTx (CalBiochem) bioinfo.vanderbilt.edu/webgestalt/) to identify GO category and KEGG path- was performed for 90 min at 37 °C, 5% CO2. After incubation, cells were ways that are significantly overrepresented, using default settings (the top 10 washed twice in PBS solution and processed for the chemotaxis assay. Cham- pathways with the most significant P values found by a hypergeometric sta- bers were incubated for 21 h at 37 °C, 5% CO2,toallowmigrationtooccur. tistical test were identified). For GO analysis, the significance level for the After incubation, nonmigrated cells were scraped and filters were fixed, adjusted P value was set at 0.01 and the number of minimum genes for stained with H&E (Diff-Quik staining protocol; Medion Diagnostics), and a category was set at 2. For KEGG analysis, the top 10 results are shown, and mounted on slides (7.5 μg/mL DAPI in Vectashield mounting medium). DAPI- the minimum genes for a category was set at 2. Microarray data were de- stained nuclei were counted, and relative migration values were posited in the National Center for Biotechnology Information Gene Expres- calculated. As negative controls, the only protein storage buffers diluted sion Omnibus repository and are available under accession no. GSE43029. in serum-free RPMI 1640 medium were used. Statistical analysis was per- formed by using two-sided Student t test for unpaired data. Gene Expression Assays by Real-Time qPCR. Total RNA was extracted from tumor samples with TriFast reagent (Euroclone) and reverse transcribed with ACKNOWLEDGMENTS. This work was supported by Federico Ghidoni memo- random examers. Primers for real-time qPCR were selected within the same rial funds (to F.A. and R.T.) and a grant from Fondazione Cariplo (to R.T.).

1. Jemal A, Siegel R, Ward E, et al. (2008) Cancer statistics, 2008. CA Cancer J Clin 58(2):71–96. 14. Patel S, et al. (2012) RNASET2—an autoantigen in anaplastic large cell lymphoma 2. Auersperg N, Maines-Bandiera SL, Dyck HG, Kruk PA (1994) Characterization of cul- identified by protein array analysis. J Proteomics 75(17):5279–5292. tured human ovarian surface epithelial cells: Phenotypic plasticity and premalignant 15. Acquati F, et al. (2011) Microenvironmental control of malignancy exerted by RNASET2, changes. Lab Invest 71(4):510–518. a widely conserved extracellular RNase. Proc Natl Acad Sci USA 108(3):1104–1109. 3. Feeley KM, Wells M (2001) Precursor lesions of ovarian epithelial malignancy. Histo- 16. Luhtala N, Parker R (2010) T2 Family ribonucleases: Ancient enzymes with diverse pathology 38(2):87–95. roles. Trends Biochem Sci 35(5):253–259. 4. Kurman RJ, Shih IeM (2011) Molecular pathogenesis and extraovarian origin of epi- 17. Shaw TJ, Senterman MK, Dawson K, Crane CA, Vanderhyden BC (2004) Character- thelial ovarian cancer—shifting the paradigm. Hum Pathol 42(7):918–931. ization of intraperitoneal, orthotopic, and metastatic xenograft models of human 5. Kim J, et al. (2012) High-grade serous ovarian cancer arises from fallopian tube in ovarian cancer. Mol Ther 10(6):1032–1042. a mouse model. Proc Natl Acad Sci USA 109(10):3921–3926. 18. Sica A, et al. (2008) Macrophage polarization in tumour progression. Semin Cancer 6. Bissell MJ, Hines WC (2011) Why don’t we get more cancer? A proposed role of the Biol 18(5):349–355. microenvironment in restraining cancer progression. Nat Med 17(3):320–329. 19. Murphy PM (1994) The molecular biology of leukocyte chemoattractant receptors. 7. Klein G (2009) Toward a genetics of cancer resistance. Proc Natl Acad Sci USA 106(3): Annu Rev Immunol 12:593–633. 859–863. 20. Verreck FAW, et al. (2004) Human IL-23-producing type 1 macrophages promote but 8. Klein G, Imreh S, Zabarovsky ER (2007) Why do we not all die of cancer at an early IL-10-producing type 2 macrophages subvert immunity to (myco)bacteria. Proc Natl age? Adv Cancer Res 98:1–16. Acad Sci USA 101(13):4560–4565. 9. Liotta LA, Kohn EC (2001) The microenvironment of the tumour-host interface. 21. Everts B, et al. (2009) Omega-1, a glycoprotein secreted by Schistosoma mansoni eggs, Nature 411(6835):375–379. drives Th2 responses. J Exp Med 206(8):1673–1680. 10. Acquati F, et al. (2001) Cloning and characterization of a senescence inducing and 22. Steinfelder S, et al. (2009) The major component in schistosome eggs responsible for class II tumor suppressor gene in ovarian carcinoma at region 6q27. conditioning dendritic cells for Th2 polarization is a T2 ribonuclease (omega-1). J Exp Oncogene 20(8):980–988. Med 206(8):1681–1690. 11. Acquati F, et al. (2005) Tumor and metastasis suppression by the human RNASET2 23. Polakowski N, Han H, Lemasson I (2011) Direct inhibition of RNAse T2 expression by gene. Int J Oncol 26(5):1159–1168. the HTLV-1 viral protein Tax. Viruses 3(8):1485–1500. 12. Monti L, et al. (2008) RNASET2 as a tumor antagonizing gene in a melanoma cancer 24. Chan JK, et al. (2012) Alarmins: Awaiting a clinical response. J Clin Invest 122(8): model. Oncol Res 17(2):69–74. 2711–2719. 13. Acquati F, et al. (2011) Molecular signature induced by RNASET2, a tumor antago- 25. Makarov AA, Kolchinsky A, Ilinskaya ON (2008) Binase and other microbial RNases as nizing gene, in ovarian cancer cells. Oncotarget 2(6):477–484. potential anticancer agents. Bioessays 30(8):781–790.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1222079110 Acquati et al. Downloaded by guest on September 24, 2021